What cost years in 2014 costs weeks in 2019 and in 2021 probably will work from the shelf. Are you sure you keep up?
There is a number of points in AI systems that needs expertise in AI, to keep them clear:

- Risks
- Budgets
- Timelines
- Competitive solutions
- Metrics and data culture
- Relations with state-of-the art
- Frameworks
- Vendor locks
- Human in the loop costs and involvement
- Dataset usage
- Support costs
- Interpretability
- Expandability

There is a simple way to check the sanity of what you are doing or planning to do to integrate ML and data analysis solutions. And we can do it either generally, either in-depth with several iterations depending on the size of your problems